A System-on-a-Chip Implementation for Embedded Real-Time Model Predictive Control

Abstract

This paper presents a hardware architecture for embedded real-time model predictive control (MPC). The computational cost of an MPC problem, which relies on the solution of an optimization problem at every time step, is dominated by operations on real matrices. In order to design an efficient and low-cost application-specific processor, we analyze the computational cost of MPC, and we propose a limited-resource host processor to be connected with an application-specific matrix coprocessor. The coprocessor uses a 16-b logarithmic number system arithmetic unit, which is designed using cotransformation, to carry out the required arithmetic operations. The proposed architecture is implemented by means of a hardware description language and then prototyped and emulated on a field-programmable gate array. Results on computation time and architecture area are presented and analyzed, and the functionality of the proposed architecture is verified using two case studies: a linear problem of a rotating antenna and a nonlinear glucose-regulation problem. The proposed MPC architecture yields a small-in-size and energy-efficient implementation that is capable of solving the aforementioned problems on the order of milliseconds, and we compare its performance and area requirements with other MPC designs that have appeared in the literature.Index Terms-Embedded systems, logarithmic number system (LNS), Microdevice, model predictive control (MPC), optimal control, system-on-chip.

Efficient embedded architectures for fast-charge model predictive controller for battery cell management in electric vehicles

“…However, we extended our investigation and selected few existing works that had slightly closer traits to our proposed embedded designs. These designs are discussed and analyzed as follows: A closely related work was presented in [35], which proposed a hardware-software co-design design for MPC. This design comprised a microprocessor and a matrix co-processor.…”

“…The design utilized a logarithmic number system (LNS) instead of a floating point, and a Newton's algorithm instead of a HQP, as in our design. Unlike our design, in [35], the model parameters were pre-calculated offline and stored in the microprocessor. In [8], an MPC-dedicated processor was proposed, which utilized a mix of fixed-point and floating-point numbers.…”

Implementation of linear model predictive control using a field-programmable gate array

“…Importantly, for the simply bounded QP problem in (3) it is straighforward to compute an initial primal feasible point, which reduces algorithm complexity. A similar approach is presented in [23] where an inverse barrier is employed instead of the logarithmic barrier used here. From a theoretical perspective, the inverse barrier is not known to lead to a polynomial-time algorithm, whereas the logarithmic-barrier is [17].…”

“…This approach takes advantage of recent advantages in the availability of cheap high density computer memory, capable of storing the potentially large number of affine controllers and state regions [9,24]. Another line of research has examined how advances in processor speed and/or the potential for implementation of custom designed FPGA-based compute platforms may be exploited [4,8,11,12,23,24]. The current paper is a contribution to this latter line of research and employs an online method for solving the MPC optimisation problem.…”

Efficient Embedded Model Predictive Vibration Control via Convex Lifting

“…This is where the efficiency-boosting achievements of control theory in the field of MPC come to the foreground: these developments allow one to implement better control methods with less resources. Because of the improvements in algorithm efficiency, model predictive control can now be implemented on embedded hardware such as MCUs [7], [8], programmable logic controllers (PLC) [9]- [11], or field programmable gate arrays (FPGA) [4], [12], [13], etc. Efficiency improvements in nominal or deterministic MPC can be divided into two main categories [14].…”